To function correctly in the presence of an IOMMU, the DMA buffers must be
mapped using the DMA channel's device instead of the MSIOF platform
device's device.
Signed-off-by: NGeert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: NMark Brown <broonie@linaro.org>

As typically a shmobile SoC has less DMA channels than devices that can use
DMA, we may want to prioritize access to the DMA channels in the future.
This means that dmaengine_prep_slave_single() may start failing
arbitrarily.

Handle dmaengine_prep_slave_single() failures gracefully by falling back to
PIO. This requires moving DMA-specific configuration of the MSIOF device
after the call(s) to dmaengine_prep_slave_single().
Signed-off-by: NGeert Uytterhoeven <geert+renesas@glider.be>
Acked-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: NMark Brown <broonie@linaro.org>

Add DMA support to the MSIOF driver using platform data.

As MSIOF DMA is limited to 32-bit words (requiring byte/wordswapping for
smaller wordsizes), and the group length is limited to 256 words, DMA is
performed on two fixed pages, allocated and mapped at driver initialization
time.

Performance figures (in Mbps) on r8a7791/koelsch at different SPI clock
frequencies for 1024-byte and 4096-byte transfers:

                   1024 bytes           4096 bytes
  -  3.25 MHz: PIO  2.1, DMA  2.6 | PIO  2.8, DMA  3.1
  -  6.5  MHz: PIO  3.2, DMA  4.4 | PIO  5.0, DMA  5.9
  - 13    MHz: PIO  4.2, DMA  6.6 | PIO  8.2, DMA 10.7
  - 26    MHz: PIO  5.9, DMA 10.4 | PIO 12.4, DMA 18.4

Note that DMA is only faster than PIO for transfers that exceed the FIFO
size (typically 64 words / 256 bytes).

Also note that large transfers (larger than the group length for DMA, or
larger than the FIFO size for PIO), should use cs-gpio (with the
appropriate pinmux setup), as the hardware chipselect will be deasserted in
between chunks.
Signed-off-by: NGeert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: NMark Brown <broonie@linaro.org>

  - Move buffer pointer and length setup to the top,
  - Make unsigned values unsigned,
  - Loop over words and increment pointers instead of recalculating them,
    which allows to kill bytes_done.
Signed-off-by: NGeert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: NMark Brown <broonie@linaro.org>

  - Add a timeout when waiting for the transfer complete interrupt,
  - If sh_msiof_spi_stop() fails, there's no need to clear IER, as the
    interrupt handler has already done that,
  - Propagate transfer failures in sh_msiof_transfer_one().
Signed-off-by: NGeert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: NMark Brown <broonie@linaro.org>

Based on an old patch by Guennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: NGeert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: NMark Brown <broonie@linaro.org>

Signed-off-by: NGeert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: NMark Brown <broonie@linaro.org>

Drop call to platform_set_drvdata() as driver data is not used
anywhere in the driver.
Signed-off-by: NAlexander Shiyan <shc_work@mail.ru>
Signed-off-by: NMark Brown <broonie@linaro.org>

The site-specific OOM messages are unnecessary, because they
duplicate the MM subsystem generic OOM message.
Signed-off-by: NJingoo Han <jg1.han@samsung.com>
Acked-by: NGeert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: NMark Brown <broonie@linaro.org>

Signed-off-by: NGeert Uytterhoeven <geert+renesas@linux-m68k.org>
Signed-off-by: NMark Brown <broonie@linaro.org>

In the implementation of __spi_validate(), spi core will set transfer
bits_per_word and max speed as spi device default if it is not set for
this transfer. So we can remove the same implementation in this driver.
Signed-off-by: NAxel Lin <axel.lin@ingics.com>
Acked-by: NGeert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: NMark Brown <broonie@linaro.org>

The only remaining feature of spi-bitbang used by this driver is the
chipselect() callback, which just does conditional GPIO.
This is handled fine by the SPI core's spi_set_cs(), hence switch the
driver to use the core message handling through our own transfer_one()
method.

As the (optional) GPIO CS is no longer deasserted at spi_master.setup()
time (through spi_bitbang_setup() and the spi_bitbang.chipselect()
callback), we now have to take care of that ourselves.

Remove the call to spi_master_put() in sh_msiof_spi_remove(), as our SPI
master is now registered using devm_spi_register_master()
(spi_bitbang_start() uses the non-managed version).
Signed-off-by: NGeert Uytterhoeven <geert+renesas@linux-m68k.org>
Acked-by: NMagnus Damm <damm@opensource.se>
Signed-off-by: NMark Brown <broonie@linaro.org>

Set bits_per_word_mask so the spi core will reject transfers that attempt
to use an unsupported bits_per_word value.
Signed-off-by: NGeert Uytterhoeven <geert+renesas@linux-m68k.org>
Acked-by: NMagnus Damm <damm@opensource.se>
Signed-off-by: NMark Brown <broonie@linaro.org>

Move clock management and pin configuration from the bitbang chipselect()
method to the SPI core prepare_message() and unprepare_message() methods.

As spi_master.{,un}prepare_message() is guaranteed to be called in
matching pairs, the clock management synchronization is no longer needed.

As sh_msiof_spi_set_pin_regs() is no longer called at spi_master.setup()
time (through spi_bitbang_setup() and the spi_bitbang.chipselect()
callback), we now have to take care of that ourselves.
Signed-off-by: NGeert Uytterhoeven <geert+renesas@linux-m68k.org>
Acked-by: NMagnus Damm <damm@opensource.se>
Signed-off-by: NMark Brown <broonie@linaro.org>

Add support for the MSIOF variant in the R-Car H2 (r8a7790) and M2
(r8a7791) SoCs.

Binding documentation:
  - Add future-proof "renesas,msiof-<soctype>" compatible values,
  - The default for "renesas,rx-fifo-size" is 256 on R-Car H2 and M2,
  - "renesas,tx-fifo-size" and "renesas,rx-fifo-size" are deprecated for
    soctype-specific bindings,
  - Add example bindings.

Implementation:
  - MSIOF on R-Car H2 and M2 requires the transmission of dummy data if
    data is being received only (cfr. "Set SICTR.TSCKE to 1" and "Write
    dummy transmission data to SITFDR" in paragraph "Transmit and Receive
    Procedures" of the Hardware User's Manual).
  - As RX depends on TX, MSIOF on R-Car H2 and M2 also lacks the RSCR
    register (Receive Clock Select Register), and some bits in the RMDR1
    (Receive Mode Register 1) and TMDR2 (Transmit Mode Register 2)
    registers.
  - Use the recently introduced SPI_MASTER_MUST_TX flag to enable support
    for dummy transmission in the SPI core, and to differentiate from other
    MSIOF implementations in code paths that need this.
  - New DT compatible values ("renesas,msiof-r8a7790" and
    "renesas,msiof-r8a7791") are added, as well as new platform device
    names ("spi_r8a7790_msiof" and "spi_r8a7791_msiof").
  - The default RX FIFO size is 256 words on R-Car H2 and M2.

This is loosely based on a set of patches from Takashi Yoshii
<takasi-y@ops.dti.ne.jp>.
Signed-off-by: NGeert Uytterhoeven <geert+renesas@linux-m68k.org>
Acked-by: NMagnus Damm <damm@opensource.se>
Signed-off-by: NMark Brown <broonie@linaro.org>

As different variants of MSIOF have different FIFO sizes, move the default
FIFO sizes to a new struct sh_msiof_chipdata, pointed to from the device
ID data.

[Moved ifdef to fix build -- broonie]
Signed-off-by: NGeert Uytterhoeven <geert+renesas@linux-m68k.org>
Acked-by: NMagnus Damm <damm@opensource.se>
Signed-off-by: NMark Brown <broonie@linaro.org>

Documentation:
  - Add missing "interrupt-parent", "#address-cells", "#size-cells", and
    "clocks" properties,
  - Add missing default values for "num-cs", "renesas,tx-fifo-size" and
    "renesas,rx-fifo-size",
  - Add a reference to the pinctrl documentation.

Implementation:
  - As "num-cs" is marked optional, provide a sensible default.
Signed-off-by: NGeert Uytterhoeven <geert+renesas@linux-m68k.org>
Acked-by: NMagnus Damm <damm@opensource.se>
Signed-off-by: NMark Brown <broonie@linaro.org>

In current implementation, CS is controlled by GPIO, which is passed
through spi->controller_data.  However, the MSIOF HW module has a function
to output CS by itself, which is already enabled and actual switch will be
done by pinmux.

Store the GPIO number in the core cs_gpio field, and ignore it if it is
an invalid (negative) GPIO number.

Loosely based on a patch from Takashi Yoshii <takasi-y@ops.dti.ne.jp>.
Signed-off-by: NGeert Uytterhoeven <geert+renesas@linux-m68k.org>
Signed-off-by: NMark Brown <broonie@linaro.org>

Signed-off-by: NGeert Uytterhoeven <geert+renesas@linux-m68k.org>
Signed-off-by: NMark Brown <broonie@linaro.org>

Both spi_transfer.speed_hz and spi_master.max_speed_hz are u32
Signed-off-by: NGeert Uytterhoeven <geert+renesas@linux-m68k.org>
Signed-off-by: NMark Brown <broonie@linaro.org>

Signed-off-by: NGeert Uytterhoeven <geert+renesas@linux-m68k.org>
Signed-off-by: NMark Brown <broonie@linaro.org>

DT doesn't instantiate SPI children if spi_master.dev.of_node is not set up
properly.
Signed-off-by: NGeert Uytterhoeven <geert+renesas@linux-m68k.org>
Signed-off-by: NMark Brown <broonie@linaro.org>

None of these files are actually using any __init type directives
and hence don't need to include <linux/init.h>.  Most are just a
left over from __devinit and __cpuinit removal, or simply due to
code getting copied from one driver to the next.
Signed-off-by: NPaul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: NMark Brown <broonie@linaro.org>

Set hardware CS(CS control function on MSIOF <-> GPIO CS) polarity
according to SPI_CS_HIGH flag on spi->mode.
Signed-off-by: NTakashi Yoshii <takasi-y@ops.dti.ne.jp>
Signed-off-by: NMark Brown <broonie@linaro.org>

Truncation on integer division in sh_msiof_spi_set_clk_regs()
results in insufficient transfer frequency (> max_speed_freq).

For example, source 52MHz, required max 6MHz
 52/6 = 8.6 --> 8, then 1/8 table selected,
and result in 52/8 = 6.5 MHz (>6MHz)

Rounding it up is a simple solution.
 52/6 = 8.6 --> 9, then 1/16 table selected,
and result in 52/16 = 3.25 MHz
Signed-off-by: NTakashi Yoshii <takasi-y@ops.dti.ne.jp>
Signed-off-by: NMark Brown <broonie@linaro.org>

Get the driver ready for the migration to the common clock framework by
calling clk_prepare() and clk_unprepare(). The calls are added in the
probe and remove handlers as the clk_enable() and clk_disable() calls
are located in atomic context and there's no callback function in
non-atomic context that can be used to prepare/unprepare the clock.
Signed-off-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: NMark Brown <broonie@linaro.org>

This simplifies error and cleanup code paths.
Signed-off-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: NMark Brown <broonie@linaro.org>

Cast pointers to uintptr_t instead of unsigned int. This fixes warnings
on platforms where pointers have a different size than int.
Signed-off-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: NMark Brown <broonie@linaro.org>

Use this new function to make code more comprehensible, since we are
reinitialzing the completion, not initializing.

[akpm@linux-foundation.org: linux-next resyncs]
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>
Acked-by: Linus Walleij <linus.walleij@linaro.org> (personally at LCE13)
Cc: Ingo Molnar <mingo@kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Use the wrapper function for retrieving the platform data instead of
accessing dev->platform_data directly.
Signed-off-by: NJingoo Han <jg1.han@samsung.com>
Signed-off-by: NMark Brown <broonie@linaro.org>

Previously the runtime PM API insisted on having callbacks for everything
but this requirement was removed a while ago so the empty callbacks can
also be removed.
Signed-off-by: NMark Brown <broonie@linaro.org>

This eliminates having an #ifdef returning NULL for the case
when OF is disabled.
Signed-off-by: NSachin Kamat <sachin.kamat@linaro.org>
Signed-off-by: NMark Brown <broonie@opensource.wolfsonmicro.com>

This adds the capability to retrieve setup data from the device tree
node. The usage of platform data is still available.
Signed-off-by: NBastian Hecht <hechtb+renesas@gmail.com>
Signed-off-by: NGrant Likely <grant.likely@secretlab.ca>

clk_get() no longer needs a character string for associating the right
clock as this is done via the device struct now.
Signed-off-by: NBastian Hecht <hechtb@gmail.com>
Signed-off-by: NGrant Likely <grant.likely@secretlab.ca>

We are clipping down the presence of module.h, since it was
everywhere.  If you really need it, you better call it out,
as per this changeset.
Signed-off-by: NPaul Gortmaker <paul.gortmaker@windriver.com>

For simple modules that contain a single platform_driver without any
additional setup code then ends up being a block of duplicated
boilerplate.  This patch adds a new macro, module_platform_driver(),
which replaces the module_init()/module_exit() registrations with
template functions.
Signed-off-by: NGrant Likely <grant.likely@secretlab.ca>
Acked-by: NGreg Kroah-Hartman <gregkh@suse.de>
Reviewed-by: NMagnus Damm <magnus.damm@gmail.com>
Reviewed-by: NMark Brown <broonie@opensource.wolfsonmicro.com>
Reviewed-by: NStephen Boyd <sboyd@codeaurora.org>

Since commit [e58aa3d2: genirq: Run irq handlers with interrupts disabled],
We run all interrupt handlers with interrupts disabled
and we even check and yell when an interrupt handler
returns with interrupts enabled (see commit [b738a50a:
genirq: Warn when handler enables interrupts]).

So now this flag is a NOOP and can be removed.
Signed-off-by: NYong Zhang <yong.zhang0@gmail.com>
Signed-off-by: NGrant Likely <grant.likely@secretlab.ca>

Sort the SPI makefile and enforce the naming convention spi_*.c for
spi drivers.

This change also rolls the contents of atmel_spi.h into the .c file
since there is only one user of that particular include file.

v2: - Use 'spi-' prefix instead of 'spi_' to match what seems to be
      be the predominant pattern for subsystem prefixes.
    - Clean up filenames in Kconfig and header comment blocks
Signed-off-by: NGrant Likely <grant.likely@secretlab.ca>
Acked-by: NWolfram Sang <w.sang@pengutronix.de>
Acked-by: NLinus Walleij <linus.walleij@linaro.org>

Instead of sending data 8 bits at a time in 8-bit SPI mode, swap bytes
and send and receive them 32 bits at a time. Tested with an SD-card,
with which this patch reduced the number of interrupts by 50%, when
reading 5MiB of data (there are also small service packets, the number
of interrupts, produced by 512-byte sectors should, of course, drop by
75%), and improved throughput by more than 40%.
Signed-off-by: NGuennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: NGrant Likely <grant.likely@secretlab.ca>

1. sort headers alphabetically
2. use fixed-size types u8, u16, u32 for register values and transferred data
3. simplify some arithmetic operations
Signed-off-by: NGuennadi Liakhovetski <g.liakhovetski@gmx.de>
[grant.likely@secretlab.ca: removed label indentation change]
Signed-off-by: NGrant Likely <grant.likely@secretlab.ca>